Reconstructing asteroids with multimodal data
Mikko Kaasalainen and Matti Viikinkoski (Tampere University of Technology, Finland)

Due to their small size and longe distance from us, most asteroids are only observable as points of light (photometric observations). We have shown that photometric data uniquely determine convex shapes. For thousands of targets, various degrees of disk-resolved data can be used for more detailed nonconvex shape reconstruction. For this end, we introduce ADAM, the All-Data Asteroid Modelling method. ADAM is simple and universal since it handles all disk-resolved data types (adaptive optics or other images, interferometry, and range-Doppler radar data) in the same manner via the Fourier transform once the 3D surface of the target is prepared for fast ray-tracing. This also enables fast convergence with gradient-based optimization. We have derived uniqueness results for the various data types.

In the reconstruction process, the difference between each data type is essentiallly the definition of the particular generalized projection from 3D onto a 2D image plane. The disk-integrated data (photometry) are just a special case of this (a sum over the surface). Stellar occultation timings can be included as sparse silhouettes, and thermal infrared data are efficiently handled with an approximate diffusion-equation solution that is sufficient in practice due to the dominance of the high-contrast (boundary) pixels over the low-contrast (interior) ones. This is of particular importance to the raw interferometric data that can be directly handled by ADAM without having to construct the usual image. We study the reliability of the inversion by using the independent shape supports of function series and control-point subdivision surfaces.